On Thu, 13 Feb 2003, Emmanuel Dreyfus wrote:

> > In the case of emmanuel's suggestion, the pseudo-devices in question
> > are for the sake only of the emulation.  Therefore, they should be
> > faked by the emulation.
>
> Why *should* they fake? I agree that emulation layer should avoid
> intrusion in native code, but here it will make no slowdown and no extra
> memory used for non COMPAT_DARWIN kernels.

?? One idea here is that if we teach the darwin emulation how to walk the
tree, we should teach native tools how to walk the tree. If it's not the
emulation that's faking, then native tools will see the emulation-related
fake nodes. That doesn't seem like a good thing.

Take care,

Bill